This is the second trade in a small experiment. I give Claude a capped pot of money, it does the research and makes a concrete call, and I decide whether to act. Everything here was written before the outcome is known. The follow-ups at the bottom are dated and added later; nothing above them gets rewritten.
Trade summary
| Action | Buy |
| Ticker | GM, General Motors, NYSE |
| Quantity | 15 shares |
| Fill price | USD 76.87 (limit was USD 77.00) |
| Cost | USD 1,153.05 plus USD 1.09 fees; IBKR shows an average price of USD 76.94 all in |
| Executed | 21 October 2026, 21:53 SGT (09:53 ET) |
| Stop order | Sell 15 at USD 74.50 stop, GTC, attached to the buy ticket, showing under Open Orders |
| Profit target | Price alert at USD 84.80, no sell order |
| Status | Open, unrealised |
Why I am doing this
One trade proves nothing. I don’t want to look at Intel alone and say AI is right or wrong, or that I made the right or wrong decision. That needs a run of trades, taken the same way, with the same rules. Hence this one.
GM had just reported a good quarter and the market sold it off hard. Claude’s call was green-amber and the numbers made sense to me, so I took it. The ticker, the size, the price, the stop and the target were Claude’s. The decision was mine.
What I believed before the trade
I believed the drop was a reaction to guidance tone, not to the business. Revenue and cash flow went up, the buyback is big, and there is no earnings report until late January. If the stock holds the level it fell to, I get a quiet few weeks to make 10%. If it doesn’t, the stop is tight and I’m out small.
I also knew I was breaking my own entry rule. The rule says buy on a close above the previous day’s high. GM had not done that. Claude flagged it and sized the call down. I went ahead anyway.
The evidence available at the time
The case for
- 20 Oct 2026: Q3 revenue USD 50.0B against about USD 48.8B expected. Adjusted EPS USD 3.70 against USD 3.69. Adjusted automotive free cash flow USD 5.9B, up 40% on the year.
- 20 Oct 2026: full-year adjusted EPS guidance raised to USD 12.75 to 13.75 (from 12.00 to 14.00). Free cash flow guidance raised to USD 11.5 to 12.5B (from 9.5 to 11.5B).
- 20 Oct 2026: new USD 20B multi-year buyback authorised, after finishing the USD 6B programme. Share count down to 849M from 904M at end 2025.
- 20 Oct 2026: shares fell 6.8% to USD 77.09 on 24.6M shares, about three times normal volume. 21 Oct close USD 76.11. The sell-off landed at a level that had held before, which was the basis for the entry near USD 77.
- 21 Oct 2026: forward P/E about 5.4, dividend yield about 0.95%.
- Next earnings 27 January 2027, outside the three to four week window for this trade.
The case against
- 20 Oct 2026: the EPS beat was one cent. The operating-profit guidance midpoint stayed at USD 15B. GM flagged about USD 2B of commodity inflation. Coverage put the drop down to 2027 uncertainty: tariffs, pickup launches, EV economics.
- 21 Oct 2026: the entry trigger did not fire. A close above the 20 Oct high of USD 78.97 was needed; the close was USD 76.11. Buying the day after a 7% drop is the definition of catching a falling knife.
- 21 Oct 2026: the stock was about 17% below its 52-week high of USD 91.8. A stock that just fell through one level can fall through the next.
- The stop at USD 74.50 is 3.1% below the fill. The 20 Oct trading range alone was 5%. A normal bad day can take me out before the idea gets a chance.
What Claude suggested
Call made 21 October 2026, about 21:45 SGT, before I placed the order.
- Buy GM, 16 shares, limit USD 77.00. Do not chase above USD 78.00.
- Stop USD 74.50, GTC, attached to the buy ticket.
- Price alert at USD 84.80 (about +10%). Hold three to four weeks.
- Reason: a beat and a USD 20B buyback sold off 6.8% to support near USD 77, P/E under 6, no earnings until late January.
- Flagged: the trigger rule had not fired, so this was a rule-break and sized at half.
- Verdict: green-amber.
What I decided
- Followed the call. Bought 15 shares instead of 16, which was what the cash in the AI pot covered.
- Limit USD 77.00, filled at USD 76.87.
- Attached the stop on the buy ticket for the first time, instead of placing it the next day as I did with Intel.
- Alert at USD 84.80, no sell order. The exit is my decision.
- Entry rule broken knowingly. Logged, moving on.
One trade proves nothing.
How it was executed
Buy ticket in IBKR with a limit of USD 77.00 and the stop-loss attached using Attach Orders, GTC. The buy filled at USD 76.87, a little better than the limit. The stop went live with the fill and shows under Open Orders as “Sell 15 GM Stop 74.50, GTC”.
At the time of the screenshot, after hours on 21 October, the stock was at USD 75.89 and the position showed an unrealised loss of USD 15.79, or 1.4%.
What would make this wrong
- A close below USD 74.50. The stop takes me out for a loss of about USD 36 plus fees. That is the plan working, not failing, but it still means the entry was wrong.
- A close below the 20 October low of USD 75.08 before any bounce. That would say the “support near 77” idea was wishful.
- The market deciding the sell-off was about 2027 and not about the quarter. Then there is no reason for a re-rate inside four weeks and the trade just drifts.
- A tariff headline. GM is exposed and the stop is tight.
If this fails, the lesson is about the entry rule, not about GM.
What happened next
- 21 Oct 2026, after hours: USD 75.89, unrealised loss USD 15.79 (1.4%). Stop showing under Open Orders.
Closed on 30 October 2026, see Testing AI trading: GM, sold at target.
Added 31 October 2026. I now add a Basic TA block to every trade post. This one uses only daily bars up to the last close before I bought.
Basic TA at entry
GM, as of the 20 October 2026 close, from IBKR daily bars.
| Indicator | Value | Read |
|---|---|---|
| Price | USD 77.09 | Close before I bought |
| MA20 | 81.39 | Price below: bearish |
| MA50 | 84.04 | Price below: bearish |
| RSI 14 | 37.1 | Weak, not oversold |
| Trend | Two down days, 83.96 to 77.09 (down 6.8% on 20 Oct) | Sharp short-term drop |
| Support | 75.07 (20 Oct low) | First floor about 2.6% below |
| Resistance | 81.39 (MA20), then 83.8 to 85.8 | Next ceiling 5.6% above |
| Volume | 15.8M vs 3.9M average | 4 times average, heavy selling |
| Trigger | 77.09 vs previous high 83.82 | Not fired |
Overall read: sharp drop on heavy volume, below both averages, trigger not fired.
Sources
- GM Q3 2026 results and guidance, reported 20 Oct 2026: INDmoney summary, https://www.indmoney.com/blog/us-stocks/general-motors-q3-earnings-analysis (read 22 Oct 2026)
- USD 20B buyback and next earnings date, 20 Oct 2026: ad-hoc-news, https://www.ad-hoc-news.de/boerse/news/corporate-news/general-motors-approves-usd-20-billion-buyback-general-motors-stock-is-worth-usd-67-6-billion/70382073 (read 22 Oct 2026)
- Daily prices 16 to 21 Oct 2026 and valuation: StockAnalysis, https://stockanalysis.com/stocks/gm/history/ and https://stockanalysis.com/stocks/gm/ (read 22 Oct 2026)
- Fill, stop and position: IBKR screenshot, 21 Oct 2026, 22:24 ET
Updates to this post
| Date | What changed |
|---|---|
| Added a Basic TA block (data as of the 20 Oct close, before the buy) | |
| Added the closing line linking to the exit post |
This is a personal record of a small, capped experiment, not advice. Claude is not a financial advisor and neither am I. Prices and figures are as of the dates shown and may be wrong; check them yourself before doing anything with your own money.